SCAFFOLD-Enhanced Federated Edge Intelligence for Urban Traffic Signal Optimisation under Non-IID Mobility Conditions

Abstract

This archive contains the complete reproducibility package and supplementary data for the research manuscript titled "SCAFFOLD-Enhanced Federated Edge Intelligence for Urban Traffic Signal Optimisation under Non-IID Mobility Conditions". This package provides all necessary configuration files, input data, and parameter specifications required to independently replicate, verify, and extend the simulation and network performance results presented in the paper. 📂 Package Contents Origin-Destination (OD) Matrices (.csv / .dat): Spatial trip generation and distribution datasets used to model traffic demand across network nodes. Traffic Signal Phase Plans: Intersection control logic, cycle lengths, and phase split configurations for signalized intersections. Turning Probabilities: Node- and approach-level routing probability matrices defining directional vehicle splits. Seed Values & RNG Configuration: Fixed random seed sets and stream configurations ensuring deterministic and repeatable simulation runs across all evaluated scenarios. Scaling Parameters: Numerical multipliers and spatial transformation rules applied for demand variation and congestion sensitivity analysis. $\alpha / \beta$ Weights: Parameter values and calibration data for multi-objective optimisation and hybrid cost functions. OMNeT++ Communication Configuration (omnetpp.ini, .ned): Complete simulation configuration, network topology description files, and radio layer parameter settings for vehicular and communication network modelling.Execution Prerequisites & Quick Start Software Requirements: Traffic Simulator (e.g., SUMO / VISSIM / TraCI interface as applicable) OMNeT++ (v5.x / v6.x) with INET framework / VEINS framework Replication Steps: Extract the contents of the archive to your local working directory. Load the OMNeT++ workspace and reference the provided omnetpp.ini configuration files. Execute the simulation scripts using the specified seed lists to regenerate the performance metrics and output figures. Citation If you use this dataset or code archive in your research, please cite the associated manuscript: